[
https://issues.apache.org/jira/browse/HDFS-6774?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lei (Eddy) Xu updated HDFS-6774:
--------------------------------
Attachment: HDFS-6774.002.patch
Hi, [~atm] Thanks for the review.
bq. The change in BlockPoolSlice - was that just a separate bug? Or why was
that necessary?
It is indeed a separate bug. Should I open a new JIRA for this?
bq. I see the code where we remove the replica info from the replica map, but
do we not also need to do something similar in the event that the replica is
currently referenced in the BlockScanner or DirectoryScanner data structures?
It could be that we don't, but I wanted to check with you to see if you've
considered this case.
Thanks for finding this bug. {{BlockScanner}} needs to remove the blocks and
the corresponding patch is attached.
I think that {{DirectoryScanner}} does not need to change, since it uses
{{FsDatasetImpl}} as input source for each scan and performs
{{DirectoryScanner#isValid()}}, which verifies whether the volume is still
available, when generate disk reports. This logic is the same as handling disk
failures. Therefore, {{DirectoryScanner}}'s consistency can be provided by
{{FsDatasetImpl}}. What is your opinion about this?
> Make FsDataset and DataStore support removing volumes.
> ------------------------------------------------------
>
> Key: HDFS-6774
> URL: https://issues.apache.org/jira/browse/HDFS-6774
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: datanode
> Affects Versions: 2.4.1
> Reporter: Lei (Eddy) Xu
> Assignee: Lei (Eddy) Xu
> Attachments: HDFS-6774.000.patch, HDFS-6774.001.patch,
> HDFS-6774.002.patch
>
>
> Managing volumes on DataNode includes decommissioning an active volume
> without restarting DataNode.
> This task adds support to remove volumes from {{DataStorage}} and
> {{BlockPoolSliceStorage}} dynamically.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
